In this episode, we are joined by Duncan Lowder and Joe Selvik, co-founders of FixturFab, to learn about how to develop a PCBA functional test system. FixturFab is a software design platform that uses generative Computer-Aided Design (CAD) technologies to automate the design process of PCBA Functional Test Fixtures. They can assist with Mechanical Fixture, the Test Instrumentation, and Test Software for both small prototype runs and mass production.
Show links:
FixturFab Blog
Flying Probe Testers
In-Circuit Tester
Wire Bonding Machine
Ingun Test Fixture
pickplacepodcast.com
In this episode, we are joined by Duncan Lowder and Joe Selvik, co-founders of FixturFab, to learn about how to develop a PCBA functional test system. FixturFab is a software design platform that uses generative Computer-Aided Design (CAD) technologies to automate the design process of PCBA Functional Test Fixtures. They can assist with Mechanical Fixture, the Test Instrumentation, and Test Software for both small prototype runs and mass production.
Show links:
FixturFab Blog
Flying Probe Testers
In-Circuit Tester
Wire Bonding Machine
Ingun Test Fixture
pickplacepodcast.com
Welcome to the pick place podcast, a show where we talk about electronics, manufacturing and everything related to getting a circuit board into the world. This is Chris Denney with Worthington. And this is Melissa Hough with CircuitHub. Welcome back, Melissa. Welcome back, Chris, Good to be back. good to be back and great job guest hosting the hack chat yesterday. I saw you got to talk to some people from your favorite place in the world, Melbourne melbourne Australia for the win. Ah, God, I love Melbourne. Great hot pot. Chinese food. Absolutely. Fantastic. Yeah. Make your way down there and get some delicious hot pot. So for anyone that's coming from the hack chat. And that just found out about the podcast. Welcome to the show. We hope you enjoy it here. Yeah. Yeah, absolutely. That was a really cool experience. That was a lot of fun. I would love to do something like that again. The great thing about that is there's this audience of tens of thousands of people that know hack a day. And so they can put together this sort of thing, and you can get an audience of, you know, 50 people and have these great conversations. If I was to try to set something up like that, it would be like two people would sign in and, but it was a lot of fun to just like answer these sorts of questions. Maybe we'll maybe once we get to 10,000 downloads per episode, we can do something like that. We have a little bit of time to prepare then. A little bit of time to prepare for that nowhere near 10,000. Absolutely. Oh yeah. Yeah. Good stuff though. But it's been a busy couple of weeks while actually, I guess it's only been a week since we last recorded, but it's been busy as mentioned, we bought our new pick and place machines and the preparation process for that is nuts. Just even like, okay, you gotta get power, right? Like you got to find an electrician and you got to see if your panel has enough room, which it doesn't, which is great. So now I've got to figure that out. That's all good stuff. And in the meantime, just trying to keep up with demand and trying to keep up with part shortages. It is nuts, absolutely nuts. It doesn't seem like it's getting better, unfortunately, We were buying this analog devices part for our customer for like 10 bucks a piece. And of course they're like all out of stock, but we found somebody who has a stock. Melissa, please guess I want you to guess how much this distributor wanted for this analog devices part, which we would normally pay $10 for. We needed 150 of them hundred dollars, more, really? A thousand dollars. more. That's insane. They wanted $1,600 per part. Oh my God. That's madness. It feels like it should be illegal. I don't know that just absolutely absurd, but they're already there. You have it. I guess if people are willing to pay for it, they'll pay for it. But needless to say our customer was not willing to pay for it. So we'll have to come up with another means of helping them get those parts or something. that does remind me during that episode a few weeks, back on Chinese new year and the effects, I know you mentioned, oh, perhaps we might not even see that big of a disruption because of the parts. And I think that's kind of what actually ended up happening. Yeah. It felt painless so weird. It was so weird. It's normally the complete opposite was a matter of fact, I noticed on our internal slack, somebody said reached out to our PCB supplier because we are a piece of PCB suppliers are on our slack. And he said, Hey, you know, could you expedite this PCB order? This customer needs these boards urgently. And I sent a quick message to the account manager. And I said, are you sure we have parts? She's like, oh, good point. Let me check on that. Why, why expedite PCBs when you don't even have parts of it on the board? Yep. So that's probably part of the reason that you hardly felt the impact of Chinese new year, Something that we've been Kind of putting off for awhile is talking about testing and we get a few people asking us every now and then to have an episode on testing. And honestly, it's because I just don't know enough about it. I've done enough of it to kind of sound like I know what I'm talking about, but I really don't. I couldn't really develop a test routine or a test fixture very well. I'm pretty weak at all this. And there's like a whole world of flying probe and in circuit tests and all these things that like, I don't understand at all. But we've been working with this company to put together a test fixtures for some of our customers. And they're great. Guys are really knowledgeable and just like we did with like our board wash show where we brought on somebody who had a lot of experience in circuit board cleaning to educate us and educate the listeners we thought we'd do the same thing with testing. So we got a couple of guys from the company called fixture fab. We have a Dunkin Lauder and Joe salvage. I really hope I didn't butcher those pronunciations, but welcome to the show. Welcome guys. Thanks for having us. Absolutely. So, well, first of all, maybe, maybe we'll start with well, either one of you feel free to give us a little bit of a rundown of what fixture fab is, what you guys do. And then also we would love to hear your backgrounds by all means, take it away either one. So at fixturfab uh, we provide functional circuit board test solutions. So anything from just building a mechanical bed of nails, fixture to a completely automated turnkey system. We can provide everything in between. So, you know, the physical mechanical fixture, the instrumentation. that does all the measurements and programming as well as all of the, uh, automated test software. Kind of how we got started in this space is I originally went to school for electrical engineering and was doing kind of embedded software development for a couple of startups and, uh, we've needed test fixtures for manufacturing. So that's kind of the world that I dove into. Before you got started with that, did you have anybody guiding you with the test fixture development or are you just trying to figure it out? Okay. I was just trying to figure it out. We were like, okay, like we're building a board in China. Like they keep sending us broken boards. How do we fix it? Fix it. Yeah. My first time building a, a test fixture it was for a company called Sphero. They did like robotic toys. one of their big ones was a BB. definitely heard of Sarah. Yeah. Yeah. They, they still do like a bunch of educational robots that are used to teach kids how to program. But it was working on one of their products where I learned how to do all of the test fixture stuff. I did work with one guy there. This guy Ryan Edwards, and he like, didn't go to engineering school or anything, but he was just like a hacker and loved to build things. He figured out how to do the test fixtures originally. And then he kind of like mentored me a little bit and then I kind of took it from there. So it's, it's a fascinating world because you have to like build a mechanical thing, while integrating electronics and all the software with like it's, I found it fascinating. That's kind of why I had stuck with it. The mechanical part. I feel like for many of us, and when I say us, I mean, us in the electrical engineering world is the hard part. It's always like, well, yeah, I can figure out how to measure this and that, but how do I actually get it to do it 10,000 times in a row? Yeah, exactly. It's crazy how expensive some of these fixtures can be. Cause you know, you can spend a quarter million dollars on a test system if you want. And then like also like, you know, we were at a fairly, low cost cm, like over in China. And so their approach to testing is completely different than like what's in the us where it's, you know, these super rigid, like very robust, like metal fixtures. And over there it's like plastic test fixtures. And the idea was like, oh, we'll just build 50 tests fixtures. If one breaks, you know, we'll throw it away, bring in a new one. Yep. Yeah, Where your fixture fab started was solving the mechanical problem. I was like, you know, these fixtures, you're just putting holes in plates. Like it's not hard, but it's very time consuming. And if you put a hole in the wrong spot, like, you know, the test fixture doesn't work. so I wrote a bunch of Python scripts originally that generated the kind of 3d models for these test fixture plates. And then that was what we were originally trying to sell. It was just like a software design platform for these test fixtures. No kidding. We're still working on it and we want our customers to be able to use our software. It's just, you know, still in the pretty early stage and we use it internally. But no one was interested in buying just the software aspect. And so we were like, okay, like how can we make money? And that's kind of what got us into actually manufacturing test systems ourselves for our customers. yeah Fascinating. I appreciate you that, that background. And, and how about yourself, Joe? What what sort of your background, how did you end up with fixture fab? So back in college it was computer engineering, which has half electrical engineering, half software. I just kind of naturally went down that software path. Like a lot of people do and Duncan and I met at our first job out of college seven, eight years ago, and kind of always At Sphero? Oh no, this is at Seagate, a computer hard drive company. Oh, that's you mean that small? No, nothing. hard drive company. Yeah, They make half the computer hard drives you might've heard of them. you might've heard of them. Yeah. Duncan, I always wanted to, like, we talked about doing our own business someday and this one day he came to me, he's like, oh, I think these scripts that generates CAD files for something called the test fixture. Cool. I don't know anything about that, but I can help you in the software because I'm bored and like working on side projects. And that's where we kind of like started at least I think it was a half a year later working on fixture fab, then part-time with Duncan And eventually it got to the point where we just kind of wanted to run with it. And for me, it was really interesting, kind of going back then to my hardware or electrical engineering side of college that I never touched again since the professional world. So for me, it was really interesting writing software that generates mechanical design files, because then you're also working with mechanical design and also electrical integration. EE work also goes into that. So, all of a sudden I went from working on just software products to software electrical and mechanical work. And that blend between the three disciplines was really interesting to me. so so neither of you are mechanical engineers out of college? No, you're just picking this up. See, I would have assumed one of you is a mechanical engineer because when it comes to test fixtures, it's so mechanical. It really is It is but like Duncan kind of hinted at though, like in the end, like all you're doing is drilling some holes. It's a very precise holes in a 2d plate. And even placing just 20 of those, like accurately is really time consuming, but not that hard of a problem to solve when you have a list of coordinates and where they need to go on a plate. And maybe it's a little different drilling this, or like laser cutting this than it is like CNC machining it. But in the end, the design files are pretty much exactly the same, whether you're 3d printing it, laser cutting it or machining it. Okay. So I, I don't want to get too far ahead of ourselves because I have so many questions about everything you just said, but I also want to like, take a step back for a second. For listeners who are not familiar with what exactly we're talking about with a test fixture, can you describe kind of like, you know, theater of the mind, right? That's that's podcasting, if you think of it that way it's only verbal. So, you know, paint us a picture of what we're talking about when we say we're using a test fixture for a finished circuit board. Yeah. So, how we love to describe a bed of nails test fixture? Which in this case, whenever we say a test fixture, we mean a bed of nails test fixture, and they kind of look like a giant panini press. So it's a big panini press with a bunch of spring loaded needles in it that make contacts with a circuit board to take measurements of voltages check for short circuits and program processors. Okay. a big panini press, big expensive panini press. Yes, yes. Yeah, exactly. And far less delicious, less delicious. So I guess, like I have a lot of questions about sort of like testing itself and, and what you're able to test and how you test it and that sort of thing. But I'm before we kind of get into that, I'm wondering if, since we already kind of started, can you talk about sort of like, okay, when you say you have to drill holes, Into a plate. The first thing that I think of in my mind is I'm just going to take my drill file, my Exelon drill file out of Eagle or KiCAD, or whatever. And I'm just going to send it to a CNC shop to just drill these. Well, I guess they can't because it's gotta be insulated, right. You can't be can't drill it out of aluminum. You would just have, okay. So what do you do? Like, are you using those Excelon files? What happens next? Yeah, so like there's a couple of different ways go identifying the test point coordinates. So a drill file works. Sometimes the pick and place like coordinate file will work cause like they'll have the x,y oKiCAD here the test point component is located on the board. And then like we have a couple of different parts of these that can also work with native design files. So like KiCAD, Eagle, um, Sort of how CircuitHub works, where you don't need to do a special export. Yeah. Yeah, I was actually like, I want, I was like, I want this to work like CircuitHub, you've just upload your file and a test fixture in the mail. We're not there yet, but I'll take it. yeah. And then for the actual cartridges we have two different styles of fixtures. We have one called a development fixture, and these are for like your initial prototype boards where it's just made out of MDF and acrylic. It's good for a couple hundred units. And it's great for developing firmware or something like that. And then for kind of a production level of fixture where you're testing, you know, tens of thousands, if not hundreds of thousands of boards those are typically made out of FR4 so it's like a fiberglass composite, just like what circuit boards are made out of and that material. the same 1.6 millimeter thickness? Are they Oh, it's it's thick. So, the ones we use are typically 10 millimeters thick. Oh, wow. Oh, wow. Okay. it's a big chunk of fiberglass. Yeah. And it's very rigid and dimensionally staple. so that's kind of why it's chosen. One of the ways we're able to do this without having a mechanical engineer currently at our company is a we've buy our test fixtures from a German supplier called Ingun and then we just do the modification ourselves. well you buy the panini press from yeah, we buy the panini press. Yeah, yeah, yeah. Okay. So you're buying the panini press from this German company. But are you buying that fiberglass too? Or, or is that from another? It depends most of the time we'll just buy it from Ingun but if we need to do something special, then we'll just buy sheets of FR4 and modify it ourselves. Gotcha. Gotcha. Okay, cool. And then please tell me, you have like a super sweet, like Haas CNC machine there, or are you farming that out? We do have a CNC. If we want a Haas we're not quite there yet, but we have a Roland MDX 50, so it's a little, model maker mill but it's perfect for what we're doing But that's awesome. You're doing it yourself. I think that's super cool. You're doing it right in house. Yeah. And I mean, we got started using a shapeoko like just little, Minimum viable product, man. Just, just get, get it out the door. He can start shipping, you know, you can wait until it's perfect. We started out with a $4,000 pick and place machine. You know, you just, you get started where you can and we just cut a purchase order for nearly a million dollars for pick and place equipment. It goes up. You'll get your Haas someday. Trust me. Um, That's cool. So you You're not just designing it and then saying, okay. German company build that. You're buying sort of a raw, a raw system, an unfinished system. And you're building yourself that the finished test fixture. Yep. Yeah. So we build out the, finished a test fixture ourselves, and we have kind of a different approach from the traditional way of going about functional test systems. So. Okay. Traditionally, you kind of manually wire them. So you'll a wire out from the receptacles, which hold the test probes and the plate, and then connect them either to some sort of interface that'll connect to a big test system, like ICT tests or something like that. Or, to a big connector block, that'll connect to a instrumentation rack with a bunch of really expensive equipment in it. Our approach is like, we specialize in testing mostly small IOT devices. So, you know, the security sensors and water monitors and stuff like that. And so for all these, like you don't need a hundred thousand dollars rack of equipment. Like you just need to be able to program a processor and check a couple of voltages. To do that we use these modules from a company called acronym. They're basically like ruggedized Arduinos is kind of how I'd describe them. And so they connect via a PCIE connector that we put on a custom circuit board. And so we'll solder the receptacles directly to that circuit board, which we call a test point carrier board. And then you just plug in these modules that they hook up to a computer over USB.. And then we have a couple of Python scripts that will do all the testing. And so instead of doing all this manual wiring, it's a circuit board. I have to you threw out a lot there Yeah. And I appreciated it, but I, I understood it, but I w as, as we always like to do, we always like to try to help folks who are getting new to this and kind of break it down into little chunks. So, first of all, you talked about wire wrap. Now, when I started in this industry, I learned all about wire wrap from an IPC. It's called the A610, it's a guideline for quality. Anyway I had no idea what it was, but I had to pass a test to explain that I knew how to inspect it, but I had no idea what wire wrapping is. I could describe it, but I feel like perhaps maybe you're more familiar with it than I am because I've never actually done it myself. Could you describe what a wire wrap is? It is pretty fricking cool, but Yeah. And this is something like where I learned wire wrapping was my first electrical engineering class in college. I think like we made a clock and we wire wrapped the logic circuits together for a digital clock. I built mine into a stuffed dog. I'm like, Ugh, stuffed animal electronics, like great Perfect. yeah. So wire wrapping you take a really small gauge wire, I think what like 28 to 32 gauge or I think it's typically 30 gauge. Yeah. Yeah. And you use a tool to wrap it around a square posts, kind of like a longer version of a 0.1 inch header pin and using this tool, you can create a really tight wrap It's shaped like a pistol or something, isn't it? The tool. Yeah. It creates a really good electrical contact. They're very robust and reliable. Yeah. And I mean, there's a reason why it's still used, you know, what, 50 years, it was wire wrap before circuit boards, like before they figured out the whole etching thing and plating holes. It was all wire wrap. Yeah. Yeah. And so it's how a lot of test systems are still built to this day. And it works well. You just end up with a giant rat's nest of wires. It's a pretty manual process. There are companies who have automated wire wrap machines where it's like a robotic arm that goes around and does point to point wiring. But we're not big enough for that. That's gotta be the most bespoke industry to belong to is automated wire wrapping machines. Like you got, gotta have like only 20 customers in the world for something like that. Maybe more. I dunno, I, for all I know. It's yeah. I, I really wonder, like, who actually makes those, or if they're just kind of like custom built in houses, like, you yeah, maybe. Someone's pet project Yeah, yeah. Maybe. And then you talked about in circuit tests now this is sort of something I have in our notes here. Because most of my career I've always kind of worked in contract manufacturing for low volume production. I've never really been in an environment where we've done something at a high enough volume or perhaps a high enough reliability where you could afford these expensive tests, fixtures, but I've sort of been adjacent. So I've sort of been familiar with the terms ICT in circuit test. I've been familiar with flying probe. I've been familiar with your, your gen rads and your Teledyne's and all, but I don't know what any of that stuff is. Like I walk by it and there's a big, giant vacuum hose sticking on the back of it. And there's some engineer with a Coke bottle glasses who knows what he's doing, but I have no idea what's actually going on there. Could you maybe like, cause, cause my, like I know functional test fixtures, functional test fixtures is like the thing's actually doing its job and you're just measuring to make sure that, you know, in this not real-world environment, it's acting as if it's in the real world environment. But could you maybe talk about, do you know what ICT is and what those old GenRad machines did and what flying probe is and that sort of thing. Yeah. I like, again, I'm not very familiar with in circuit tests either. So I can cover a little bit, but not too know more than I do though. And it's, it's one of those weird things where like depending on the product that's being produced, sometimes you don't do it. Like at Sphero, all the circuit boards were so small that there weren't enough test points on it for it to be worth doing the, in circuit tests. Because the main thing with what in circuit testing checks is a, it's a unpowered test. So you're mostly measuring like resistances and capacitances between various test points. To do that, you want to have ideally one test point on every single net of the board, like of the circuit board. And so it's like, if you want So you have to design your board for in circuit tests, like that's part of the design. Yeah. That's part of the design and it's getting harder and harder as, you know, electronics get smaller and smaller, more dense components on it. It's harder to, you know, put enough test points on it and you can do fancy things like try and probe component pads and stuff like that. But it just makes the reliability of the fixture less. So like, at Sphero it's like all the circuit boards are small enough. We couldn't put enough test points on it to make iCT worth it. And so we would only do functional testing where you're powering the board and making sure you can talk to the various functional blocks on it. Yeah, but in circuit testing, it's a unpowered test where you're measuring resistances and you can kind of characterize whether or not components are correctly assembled on the board by using a golden sample basically. Okay. learns what that golden sample looks like. And then you can use that to test all the other ones. Are these flying probe machines effectively doing the same thing? They're doing an in circuit test, but rather than a dedicated test fixture they're just moving the probes around the board to different points and performing the same task. Okay. Yeah, exactly. And I'm like flying pro machines are great if you're doing a smaller volume runs. But then like once you start building, you know, thousands of boards, it's much, much faster to have a dedicated fixture because instead of having to mechanically move probes around, you just have all of the probes connected and then it can just run through all the tests really quickly. I haven't looked at the big ICT testing machines very much, but I'm pretty sure it's just a giant bank of relays where it just cycles through connections. Yeah, I was going to ask you what that must be. It's gotta be probably solid state relays of some of some sort bunch of fats or who knows, but yeah, that's that's something it's a total black box, but I'll tell you what.. Have you ever had that experience where you're driving in fog and then all of a sudden you can see a mile ahead of you because the fog like lifted off. That's how I feel having you just explained to me what in circuit test is and everything. Cause it's always been so confusing to me. Yeah, it, and it's like one of those things where it's like, you always hear this term and then I never saw it used in production until like, the job I had right before starting fixture fab. And flying probe tests for listeners, if they're not familiar with these things just look up YouTube of watching these things run. They're kind of insane looking they're really, I'll try to, maybe I'll try to find some links and, and we can include some in the show notes, but they're, they're just wild to watch they're these long needles that just, it looks like the creepiest, you know, alien you've ever seen in your life. Walking all over the circuit board to test everything. Okay, so that, I appreciate you explaining that. Cause that's something that's always confused the heck out of me. Now I had another question before we get into the meat of it. If, as if this isn't the meat of yet. What is boundary scan? I constantly hear about this as like the coolest thing ever. And it's changing the world and boundary scans the new beat, like what is boundary scan? Yeah. And this is another one of the areas where I don't have enough experience in it. Like we haven't done it physically ourselves, but we have helped a couple of customers who've been using it. and I believe what boundary scan does is it's a protocol built Into processors and various ICS. And it allows you to kind of do ICT testing, but without a fixture. And so you can take measurements and double-check that pins of various ICS are connected together as they're supposed to be. Yeah. And so it's, it's a software base test approach where it uses kind of stuff baked into the Silicon to actually run these tests. You typically buy an expensive software license and probe to do it like programmer. Yup. Yup. Yeah. And a little bit of Googling I saw that it, there's a bit to get started with it. And like you say, only certain devices support it. Not every kind of design is going to support it. You kind of have to start from whatever chips that you choose. They specifically have to support it and you follow their guidelines to be able to use it and everything. But I guess the end result is you can significantly reduce the time and expense of doing in circuit test. If you design it right, to be able to use boundary scan. Yeah, you can eliminate the need for an ICT fixture, like an insert cut test fixture. If you correctly design your board, but it's one of those things where it's like, you need to go into the engineering of your product expecting having design for tests, being one of the main requirements for it. Yes. It's one of those things where the larger companies, are definitely doing that work, but kind of like hardware startups, where they're just trying to get something working. It's usually, you know, they don't even think about test until they're trying to launch their product. Yeah. Yeah. They're like, oh wait, it works now. Now how do we get it made? And then they come to me and I'm like, you gotta, this is a terrible design. I can't build this thing. And then I T asked them to redesign it so I can build it. And they're like, okay, now we need to test it. And they go to you and they're like, this is terrible for testing. I can't test this, like this. Yeah, exactly. Yeah. So if you can approach the original, design of a board with being like, I'm going to use for boundary scan, then go for it. It's a great tool. I don't know enough about it to really talk about it in detail. but the people I know who do use it swear by it. Cool. Cool, cool. Good to know. Alright, well, I have all kinds of questions for you when it comes to what I think you are probably quite familiar with. And, and that's the functional test process because like when I think of when I think of a product you know, let's, let's just, well, okay, so you talked about water, it's some IOT water controller, right. That everybody's growing cannabis and I'm sure they're there, they're making these things like crazy for that market, you know? Presumably it's going to have a pressure sensor humidity, sensor, water detection it's going to have all kinds of sensors on it. How does one go about doing a functional test on say, water detection, sensor, right? How do you actually go about figuring out that that sensor is responding to something? Or do you, do you literally just like, have a spray gun in that? Oh yeah. There's water. Yeah. And there's a lot of approaches to this, but one of the things in like, we find people like, especially engineers, you know, engineers love making perfect things. And so When they go about designing a test fixture, you know, instead of validating that a board was manufactured correctly, like that is the purpose of a manufacturing, like functional tests fixture is like, are all the parts on there? Does the board work as it's intended to. What a lot of engineers start doing is designing a validation test fixture where it's like, they're actually validating their design. And it's like, you know, do we really care if you know, it's exactly 3.3 0 0 0 0 0 volts? Or can that be, you know, 3.2, eight bullets and like, have it be okay. So with sensors, like this is really important. Your tests could be as simple as like, Hey, it's a ITC based sensor. Let's make sure we can identify that the, you know, sensor IC is there, we can get a temperature reading from it and it's, you know, around room temperature, like, cool. That's fine. But the temperature reading is, is to, to your system. It's just a voltage, right? It's just something between zero and five volts. yeah. And so it's like, okay, if like we get something cool that's all we need to know. And then if you're not blowing a hot air gun on it just to make sure it climbs or something. You just say no, the communication is there. Solder joints are good. Moving on. Yeah. And then you can also, if the, like, you know, your temperature sensor needs to be calibrated or something like that, then you could throw in a thermistor or something like that. And build that into the test fixtures. So you can be like, okay, at this one temperature point, let's set, you know, 2.7 bolts to 57.6 degrees fahrenheit or something like that. Okay. And then the step after that is , if you need a multi-point calibration, that's when you can build in like a Peltier cooler or like a small heater or something like that. And then you can change the heat of that component. yeah, yeah. Yeah. That makes total sense. So, so let's say, okay, so you talked about a temperature sensor, but I guess you could extrapolate those same principles over almost any kind of sensor. Cause I was looking up before we were recording, I was like, okay, let's just, what kind of like sensors can you get for an Arduino? And there's we, we mentioned temperature, humidity, moisture pressure, right? Like force weight. Like if you, cause you can make a scale accelerometers there's gas sensors for detecting oxygen, carbon monoxide, light sensors microphones, obviously touch sensors, IR sensors hall effects. I mean like the list just goes on and on and on. And so each time you're given another sensor, you, you have to come up with a new test routine for that sensor. Yep, exactly. And it's kind of, you know, we just keep building up our suite of like, oh, it's a new one. How do we deal with this? was it the PIR sensor? I'm not sure what that acronym actually stands for. It's like one of the yeah, like some sort of motion sensor. Okay. Okay. And to do that testing, like we ended up putting a pinwheel or like a wheel on a servo and like put like a w I think it was like an aluminum plate on it. And so we'd spin that around and see if the motion sensor would trip. So it's, every time every different sensor you kind of have to be like, okay, like, how can we test this? And for the most part, we try and not do physical things. So it's not having to move something, um, like what, it's not a sensor, but motor, motor drivers and stuff like that. yeah, oh, okay. We need to make sure we can drive a motor. And so for that, if we need to vary the load, you can use some sort of electronic load on it. Or most of the time we just throw a resistor and inductor on there and be like, okay, let's approximate. What the motor is that it, this device will be driving. And then we can just measure the current through there to validate it. Yeah. Rather than actually spin up a motor you're just putting this fake load on it. And then presumably once you've sort of, once you've sort of figured out how you test that motion sensor, anytime you get another motion sensor, you should be able to roughly plug and play that test process to the next customer. So that's kind of cool. You're kind of building your NRA. I'm calling it NRI, but like, you know, it may cost you guys $5,000 to figure out how to, how to test a a motion sensor , but once you figure that out, you don't have to figure it out again. So the next time a customer comes to you with a motion sensor, you're not spending that same $5,000. You just figured you're spending the $500 to figure out how to integrate it into their, into their particular thing. exactly. And it's, it's probably one of the benefits of actually talking to a test fixture contractor is like, you can just get some insight into how you should approach testing a board, because you can dive as deep as you want and you're just reinventing the wheel. So sometimes it's worth, you know, you know, focus on the important things, which is what your product actually does. Yeah. You mentioned one thing already, which was like, do you need it to be 4.3 0 0 0 0 0 volts? Or can it be 4.28? And. I think generally speaking, the concept of tolerance is very important for engineers to learn. Yeah. now engineers who've been engineers for 30 years. They know this by now, but if you're just getting started, like there's a stackup of tolerances. It's okay. This is normal. You know, it's still gonna work at 4.28 volts. You're going to be okay. yep. Yeah. Cause test fixtures have tolerances too. So it's like, you know, you're measuring that voltage with something. And so it's like, how calibrated does that need to be? Cause you need to calibrate your instrument and then you have this big, you know, spring-loaded test probe that it's like the resistance of it increases over time. As it like wears out. If you really need something super accurate It's just increases the cost of maintaining that fixture the entire time it's on a factory line where it's like, you know, it could even be the difference between a 1% and a 5% measurement tolerance, on a, just, you know, regular voltage rail. Because most of the time it's like, you know, if it's plus minus 5%, you're close enough, the board was built correctly. And you know, whatever is off is probably just tolerances from the fixture, So oftentimes when you're like, okay, so you have a bed of nails, right? And by its very nature, that means that you need to have contact with the circuit board somehow. How frustrated are you when you have to contact a a solder joint that has flux on it? Does that drive you crazy? If you have to repeat it over and over. It just causes so many issues. I never knew that you're supposed to clean test probes until we had that issue where it was like probing a through hole component lead that was always covered with flux. So, to do through whole leads, you kind of use like a cup tipped probe. and, it just fills up with flux, over the course of the day. The first day we deployed the fixtures, everything's all good. And then within a week it's just constant failures and it was like, yup. They have to clean the tip. Like they use a wire brush to clean it like two or three times a day. And then they have to replace those probes, like, you know, every couple of weeks versus like six months. Yeah. Yeah. I mean, it does sound annoying, but washing is very expensive and time consuming and it, and the trade off there might be worth it that they have to clean their probes and replace their probes more often. Like, and one of our suggestions is like, okay, like throw a test point. That's not going to be covered I was just going to ask you, I was just going to ask you if you could do something like that. Yep. And it's like, sure, you're not testing like typically with a through hole components, they're mostly connectors are the ones that we see that people want to probe. And we're like, you know, sure. You're probing a test point instead of the connector that soldered on. But the reliability of that, like over the life of the fixture will, you know, more than pay for itself because realistically with connectors, like those solder joints are typically good enough. It's not a risk component for you to not pro. Yeah, we, we have we have some customers that have made their own test fixtures and they they wouldn't be embarrassed if I, if I said this, I've said it to them already, they're less than ideal, less than ideal test fixtures. And, and they, you know, they just like wires and hot glue and all this kind of stuff. And it's just not reliable. Have to be repaired all the time. So, so cleaning a little bit of a probe here and there is really nothing compared to some of the things that can be done on a, on a hack together, a test structure don't get me wrong. It works, the test structure works and it was really cheap to make, but the expense of maintaining, it might at some point be more expensive than a professionally paid test fixture. In other words, my whole point is tests, fixtures require maintenance. You can't, you know, you can't just expect to use them nonstop. And with regard to maintenance, one of the things that I find is like the most. Commonly misunderstood thing about test fixtures and those spring loaded contacts is that you are not supposed to secure the spring loaded contact. You're supposed to secure the socket for the spring loaded contact. And I can't tell you how many times we've gotten a fixture from customers where they put the spring loaded contact, solder that thing in, pressed it in, crimped it in whatever it was. It was fixed in there forever. And then you're like, great. How do I replace this? Like, what do you mean replace it? Oh no. Yup. That's So many times. receptacles, receptacles, receptacles, like, you can buy test droves from, you know, spark fund out of fruit and stuff like that. I'm actually not sure if they sell receptacles on those sites, but, you know, Maybe that's part of the problem. yeah. Go to Ali express, like, or even Amazon now has them like just search receptacles and then like the pro number, like and you'll find something, but use receptacles. Yes, please. For the love use receptacles. And what that means is basically if, if you're, if you're still not following us, it's you can pop those spring-loaded contacts out and pop a new one in, and they have a little socket that they slide into a receptacle they slide into. So you're not, you know, as that spring, cause that spring will break I've we've had spring or they get bent. So like the spring loaded contact, we'll just get a little bit of a little kink in it. And then all of a sudden it's not going up and down. Nice and smooth. You got to pull that out and replace it. I've seen that so many times, so many times. Yep. I mean, I'm guilty of it too. The first couple of tests, pictures I made I've soldered those probes straight into it. Everybody does it, man. Everybody does it. I know. It's such a common thing. Okay. So that's like with sensors, we talked a little bit about. You know, you're going to come up with a cool little thing. If you have an oxygen sensor you're going to, you know, I don't know, blow some oxygen on it or something. That'd be really dangerous. Don't do that. Just measure the ambient oxygen in the room, but let's say something that's a little more complicated than just measuring the voltage. Will you actually go through let's say you wanted to make sure that your led turned on. Do you oftentimes just see, okay the voltage is sending to the led or do you actually want to see the light being emitted from the led? Will you go, is that far to put a little light sensor on it to look for that? Yeah. Our recommended approach, we'll be like, can we just check the voltage drop across the led? Is that good enough? Um, Yeah. Cause then you're like, oh cool. It happened, but you know, if you can't make that measurement, then you can't do that. So to do optical kind of led testing, it's pretty common. And there's a couple of companies that make a dedicated led testers. So it's a little piece of fiber optic that you then Mount, either over the top of the led or if it's a right ankle led you can mount it right by it. And then with those, you can tell the color of the led, the intensity. So they're pretty cool little devices. It adds a decent amount of costs to the fixture, I think like 500 to a thousand dollars for the device, but they work But maybe you're using some expensive $50 Cree led and it has to, you know, and then you want to measure it's got enough light and all that kind of stuff. Yeah. Interesting thing about those Cree LEDs and, and there's others, Oz Ram, and there's all kinds of really bright LEDs. I was talking to when we were doing our pick and place evaluation, this was the coolest thing. I was blown away by this, a lot of premium vehicles, and I think they're coming down to lower market vehicles now are using led headlights. And when I was a kid, I remember I wanted to change out my headlights to Xenon. That was like, all the rage back in the day was Xenon headlights. And I remember everybody on forums, cause there was no Facebook. There was no any, you know, you had to go on forums and these, all the geo city sites to find people who did this kind of stuff, remember geo cities oh my gosh. Anyway. And nobody ever said don't do it. Don't do it. Because what happens is when you replace your lights you can violate your or you, you can not pass inspection because your light. It has to be in the right line cause you don't want to blind the car in front of you if you installed it wrong and your lights went up in the air you would, you would blind the driver in front of you in the rear, in the rear view mirror. And so that has to be at a certain level and everything. Well, when automotive companies started to make led headlights, they had this problem where they were, they were struggling to control the position of the headlights and get consistent headlight. And they, they were like, you know, we're placing these things perfectly. When they looked at the circuit board, they were placed absolutely perfectly on the circuit board. The problem was the dye inside the led has a tolerance. And so what kept happening was that dye would raise and lower and they'd get this weird looking headlight line. They wouldn't get a nice crisp, clean line. So now pick and place companies are actually pre inspecting the dye on the led with the fiducial camera.. Calculating an offset and then placing it offset on the circuit board so that when you light up a headlight, it's all just a perfect line. Wow. That's crazy. Isn't that cool. It is so cool. It is so cool. I was blown away by that. It obviously started in the automotive market, but there's other applications for it too. I just thought I was like, they're like, so do you need that option? It's like a $20,000 software option. I'm like nooooooo. We do not make automotive headlights. I do not want to pay. But if you are that $20,000 is totally worth it. Yeah, exactly, exactly. But like I saw my car has, has led headlights and it can like, you can put it in, like, I forget what they call it, like party mode or light show mode or something like that. And it, and it blinks all of the, all of the LEDs individually and like to the music. So like as it's playing music, it blinks it's so it's a stupid gimmick, but it's, it's fun. Anyway. But I, I, I, every time I ever, like, I bought this car and then I went to these pick and place manufacturers, I visited them and they were explaining to me what they had to do to get this to work. And now I can't help every time I walk by my car and look at them and I'm like, oh my God, that's amazing how they did that. Really. So the testing testing goes all the way up front and the, and the pick and place as well as the . Post placement. Anyway, that was a total digression. I just thought that was the coolest thing ever. Okay. So how about how about like, say, say, you know, we, we do these tests where the display it's a liquid crystal display. It's not a real fancy, you know, liquid, crystal display. It's an old school. I forgot what they're called. There's a certain term for them. Anyway, it's old school, liquid, crystal display, like a segmented display almost. And those as we're testing, those need to you, you, a human needs to read that display and confirm that that display is telling me something that, you know, oh, you know, heat zone three was on or something. I don't know, something stupid like that. Would you go as far as to like, try to somehow read that display or would you just test the heat zone three circuit that confirms that it's on? Or like, how does that stuff, how deep do you get with that kind of thing? Yeah. So again, it depends like exactly how important it is to the system. And I mean, displays are typically pretty important. One way to go about it is to get a logic capture of it, trying to drive a display or something like that, and be like, does this wave form look correct. And then, you can verify that the display actually works later at like, you know, sub assembly or final assembly stage. But you can also use computer vision. We use Python for all of our stuff, open CV, free computer vision software, and you can pretty easily capture images of what the displays supposed to look like and then kind of train an algorithm to detect whether or not that display is working correctly. Interesting. Cool. So you would just use some off the shelf, 4k camera or something and just point it at it and open CV it so cool. That's so fascinating. Yeah, it's awesome. One of the things we do struggle with a little bit is that there's typically like a potential geometer for those LCD displays to set the contrast. We haven't figured out how to automatically do that yet. There's still like an operator, twisting that pot or something like that. Yeah, I'm sure there's some kind of a robotic screwdriver or something you could do, but that would just get so expensive and complicated for an operator to just adjust it real quick. exactly. We haven't found someone want to pay us to figure it out yet. Exactly. That's exactly it. Right. Okay. What about let's say a lot of these things you mentioned there, IOT devices, how do you test that? The Bluetooth communications working, wifi communications, working USB port, cause we plug things in and we have to go into device manager and say, yup, there it is. There's the port, you know, it shows up. And then when I unplug it, it goes away and it's communicating properly. Like what, what about that kind of stuff? Yeah. So, with USB especially with USB-C and like all the different orientations that you can plug in cables it gets crazy complicated. yeah yeah, so it could be as easy as like checking. We put most of our software on Linux and so we just check the device tree and see if it shows up using a little Python script. there's a controllable USB hubs from some people you could use anything from like a what is it you have control, I think is the software package on Linux that you can use with just generic USB hubs to turn them on and off. And that allows you to identify devices. And then a company or acronym also makes programmable ones, which is great. And they have a USBC hub that allows you to test different orientations which is really fun to mess with. For kind of like wireless testing, like, Bluetooth and wifi there, those are harder. And can like, depending on how important it is, you might want to spend money on like actually getting a spectrum analyzer you know, a nice piece of equipment to validate that your radio is actually tuned and like you're broadcasting exactly where you want to be, but that's, if you're making your own wifi device, if you're just buying a wifi module. You just want to check that at works. And then for that, you know, you just have. Like USB Bluetooth module on the computer and you just check hey, can I connect to the device when it's powered? Yeah. And same thing with wifi. one thing that makes us complicated though, is doing it on a factory floor area where the 2.4 gigahertz band is notoriously just super congested. Yeah. You don't it can be hard. We we, we have, I'm trying to remember what this product was. But anyway, it had an FDDI chip for communicate over USB and. Once we plugged in, I forget if it was 256 of them or 512 of them or something like that. Eventually windows is like, I've had enough. I give up I don't know what the heck that is. Have you ever run into that scenario before? Yep. Cause you get to 255 different FTD devices and then it just freaks out like on the next Yes. it's gotta be like an integer overflow or something like that. Right. And so FDDI actually made a executable that you run and it clears it out. yup. Yeah. I remember thinking I'm like, this can't be real. Like, cause I just Googled this, like the problem I was having and then I saw this and I'm like, there's no way. Sure enough. That's what it did. Yep. I ran into the same issue. One of my work computers once. Yeah. Oh, that's the worst too. When you're like using your personal PC to do something for a customer as a favor, and then all of a sudden, like you can't plug in your mouse or something like, oh gosh. Cause everybody uses FDDI. Yup. What about yeah. Okay. So the same thing, I think I'm getting it now. So let's say, let's say you wanted to test that a speaker to being turned on. You would just measure the voltage drop across it or the resistance drop across it or something. Yeah, of course. All right. So I'm starting to get it. So you wouldn't necessarily hook up a microphone to detect a wave form. You could just say, all right, is it doing its thing? You know, we're, we're getting the same voltage we expected from our golden model, you know, or within a tolerance, right. Tolerance. We're Audio, it can be another complicated one too, if you're building a speaker, you know, like a Sonos speaker or something like that, you actually care about the audio. So if you'll do more testing there, whereas like, if it's a little buzzer or something like that, you just want to check that you're generating a voltage. Yeah. So if you're, yeah, cause if you're making an audio product, now, the tolerances really matter and now you really want to, and so you might put that in some kind of a chamber to get it, you know? Cause you don't want a pick and place machine like screaming away next to the thing. Yeah, of course. What about battery charging? How do you verify that? Like, you know, it's charging in inappropriate rate and that kind of thing. Yeah. Battery charging is actually fascinating because there's so many different modes with a kind of battery charger, ICs right now. And you typically want to make sure all three of them are cause there's what constant, current, constant voltage, and then a trickle charge usually. And so you want to make sure you can enter all three of those stages. And so to do that, we'll typically use like a electronic load or something like that, where we can program it to source a certain amount of current or have the voltage set at a certain point to exercise the battery charger in all three states. So you're not necessarily putting a battery on it. You're just putting something that's imitating the battery. Yep. Exactly. You're always trying to imitate it because with the battery in You'd fill it. environment, yeah. It's going to fill up. That's what I keep thinking. You're like all of a sudden you're going to be in trouble cause that thing's going to, and then you're going to start failing all your tests cause it can't charge anything. I remember speaking of batteries, I remember this, this sort of I don't know how to describe it. Story. I heard one time. It was, I believe it was Tony Fidel talking about it. Who's who's famous for kind of the father of the iPod. Right? I think it was Tony Fidel. And somebody said to him one time, you know, like, you know what I love about the iPod versus back then, it was like the Zune and the creative labs. I forget what they made. They said, when I buy it, it comes with a full charge. Like it's got like 90% charge and I can start listening to music right away. And he said, well, that was just a happy side effect of our testing. He says, you know, we, we decided, you know, we used to just test them real quick and then get them off the assembly line. But we wanted to do a more thorough test. And we said, well, while we're at it, why don't we just keep charging the battery the whole time? That way that our customers get a full battery. I just thought that was the neatest thing. But I, I also, at the same time, can you imagine working in an iPhone factor, an iPad factory. Batteries fail. Right. So that's kind of, that's the other thing. So have you ever destroyed some of your test fixtures or had customers destroy your test fixtures by things exploding, letting out the blue smoke? We we've let out blue smoke from at least one of our customers circuit boards, whenever we're developing a system Yes. We need multiples because you're going to blow it up on accident. Just Oh, that's a good point. Okay. So let's, we've kind of gotten into the details and nitty gritty of sort of your approach to testing, but how does how does somebody like Chris say a cm or do you normally not work with the cm? Do you normally work with my customer to get something designed for we'll start there? Who who's normally your customer. Yeah. Most of our customers are the kind of hardware company themselves. So the company who's designing the product and then, And then you send it to the cm. Yeah. Depending exactly who the customer is. If they have a close relationship with their cm, then we will also talk directly to the manufacturer. But it's typically most of our relationships with the company who's designing the product. that makes sense, because you need to get to the schematics. You need to get to the, the source code. Do you ever get the source code? Like get into the weeds there or we try not to just because it's like one more box to open up and discover all the secrets. But a lot of times, like we do need some firmware support. So like, depending on who developed the product, sometimes there's debug commands that you can send to the processor to like read sensors and stuff like that. And if those don't exist, then sometimes we need to work with the companies for more engineers so that we can have a test firmware image that we can use to exercise to the board. Right. So you're not going to look, you're not going to flash the actual shipping product. First, first, you're going to flash some kind of a test routine. And then once that all passes, then you're going to erase that. And you're going to flash the shipping product in the test fixture itself. Yeah. Yeah. And it, it's one of those things where again, like if you approach your product development with the like, oh yeah, we need to manufacture and test this device. People will typically add a debug mode or a test mode to their production firmware image where you can like pull a pin low or to send a command to like unlock the debug commands. And we like that because you only have to flash the processor once. Exactly. We do too, because it's such a pain to sit there and wait for 30 seconds for the thing to flash, you know, Yup. Time is money. And if, if you have two images, you're paying your, you're paying your cm twice the, to flash your chip rather than just once. So think about that, that really does matter. So I, cause that's what I like when you power on a device, I was just going to say, like, if it's in its shipping code, you don't necessarily know when it's going to throw a relay or, you know, read a sensor and you cause you, so you're just like, so you kind of need to have a test routine. You kind of need something so that you have predictable, this, this device is going to do what you expect it to do. Yeah, it makes it a million times easier to actually build a test fixture if you have the ability to put the processor into a known test state and tell it to do things, if you don't have that and you just have the production image, we can still develop a test, but it's not going to cover anywhere near as much as if we had a dedicated test image. So what's the sort of the, the first step then, because we, one thing you mentioned is that you actually get your customers boards when you're, when you're building these types of fixtures. So, so I'm imagining the first step is, is as a conversation, you just start out with a conversation, like what, what is your product? Do you know? And w how, how does it typically go from there? Yeah. So, kind of like, how will you first interact with customers is either they'll submit a request for quote on our website or send us an email. And what we'll ask them for is just kind of the general design files for their circuit board. We'll sign an NDA before this step so that, you know, So maybe like if you're using, if you're using Eagle, you'd send the dot PCB and the dot sch that kind of thing. Okay. For a complete. Turnkey test system we'll want the schematic. We don't need the like schematic design file, but we definitely want a PDF at least. we'll ask for manufacturing files for the boards. So, preferably like IPC 25, 81 or ODB plus plus files. Gerbers also work and then a 3d model of the board as well. That can be tricky yeah. And it's one of those. every device on the board, you know? Yeah. It doesn't need to be super accurate, but we need some sort of 3d model of the board. And we can work from that. And then once we have that information, like we can, you know, kind of review it. If the customer has developed any sort of test specification or test plan, prior to talking with us we'll ask for that as well, but it's, I don't know, 50 50 on whether or not people have actually built that out. And, from that information we can kind of like take a quick look at it and Well, let me ask, do you prefer somebody has done that already or do you prefer them to wait We prefer that they have done it themselves. you know, the person who knows the most about the product is the one designing the product, not but what I mean is like, would you prefer them talk to you early on in the process so that you can kind of coach them about, okay, when you go to do this, here's what you should be. Oh yeah. That a great question. Like, we love being contacted, while the board's still being designed or just entering the design phase of like, okay, like here are the things that you can do like, add test points on all the nets you actually want to test, make the test points as big as you possibly can and space them as far away from each other as you can. I want 10 millimeter test points. yeah, exactly. What is, what is a reasonable test? Point size is like a millimeter pretty good. Or yeah, so 0.8 millimeters is kind of like our recommended minimum. That's large enough that you're always going to hit it. You can go smaller than that, but really try to fit 0.8 millimeters. And then as far as kind of like center to center, spacing of test points we recommend a 1.9 millimeters or 75 mils or thousands. diameter of your of your test point. Yep, exactly. And then the other thing that I mean, we're guilty of not checking this when we do initial system designs all the time is checking that test points that are far away from components or like, you know, a millimeter away from a component. And if it's a tall component even farther away. So like, if you have like, a connector, you know, that's a centimeter tall or something like that, and you have a test point you know, a millimeter away from it, it's going to be really hard for us to actually Mount a probe there because you have this really long connector sticking through. Interesting. But surface Mount devices, you probably don't care too much as long as they're short surface Mount devices. Yeah. But you got a big 25 millimeter aluminum cap. Move that test point away. Yeah. Yeah. Please don't put 25 millimeter caps on your board unless you really need to, they're a pain to assemble, to put, put them in parallel, you know, just, just instead of one big one, get, get like five small ones, please. I wonder if that'd be more reliable to maybe not cause if even one of them fails, you got an issue. Nevermind. Well, okay, so this is great though. I, this is something I meant to get into. I, but I want to get into the advice thing. You've, you've given us some gold here with, with the test point sizes and everything. But before we get into even more advice, can we talk a little bit more about the process of, of getting, you know, having the communication with you? So you said the types of data you need, what, what happens next? So after that, we'll review it internally and then have a call where we're kind of discuss exactly how many of these devices are you manufacturing. Cause you know, if you're only building a thousand a year you know, having a a hundred thousand dollar test fixture might not make sense. Whereas if you're building a million than Yes, so that kind of help us kind of guide what type of fixture we would suggest. And then from there it can get into like a, a more detailed test specifications or test plan kind of conversation where we'll go through the schematic and be like, okay, like these are all of the blocks we identify and should we think should be tested and we can talk about like what the device is actually doing, because you know, every once in a while a customer will be doing something, you know, novel with a general, like a generic part where it's like, oh, you're doing this, using this for something weird. We need to know what that is, because we don't know if that it actually needs to be tested. Yeah, exactly. yeah. And then kind of from there we'll generate estimate for like, the test system. So like just a kind of general quote. And then after that stage if the customer accepts it, we'll kick off the project where we'll then kind of go into a detailed design phase where we'll talk with kind of their electrical engineers and their firmware engineers to make sure we have all of the test points needed. And the firmware support needed to kind of test the circuit board. At what point is a purchase order written? Have we already passed that point? Yeah, so a purchase order is written after we generate a estimate and a quote. And so that'll be kind of like a, this is how we think the test system is going to go together. So we'll be like, this is most of the instrumentation that's going to go with. These are the test cases that we're going to build out and the test software, and this is the size of fixture that it's going to fit in. So you'll kind of have a rough idea of how big it's going to be, what instruments are going to be in there, and exactly what's going to be tested on the circuit board. Okay. Yeah. Cool. And then then you go about making it right? You go about the actual manufacturing process, which you do yourselves, which is so cool. I can't repeat that enough. I just, I love that. Yeah. We tried outsourcing the machining initially and it's like, if we forget something or like, you know, put a test point in the wrong place or, you know, just little changes where since we have our own machine, it takes us five minutes to fix. Whereas if we had to outsource, they could take two weeks. so it's just much, much faster, plus it's awesome. Having a little CNC the way I love watching it. Yeah. Oh yeah. Oh, machines are so much fun to watch. They really are like any kind of machine is fun to watch. It's terrible, you know, trying to employ people who run machines because they're so fun to watch people just want to stand around and watch them. Sometimes. Honestly, our folks don't do that, but it can happen. You can just get sucked right. In it's like moth to a flame. They're just, they can be a lot of fun to watch sometimes. And CNC machines, especially those are just like, just so neat looking. Okay. So, so thank you for kind of walking us through the process of getting that made. Now what what advice do you have for people who, you know, design for test design for functional test fixtures? You talked about. Putting test points on your board on the nets that are important, the size of the, the test points, the spacing any other, any other advice you can give tips and tricks, so to speak? Make testing and manufacturing part of your kind of product requirements. Like at the very beginning stage of designing something it, they are just two things that are often forgot. And so like making it, part of that initial requirements space will ensure that everyone's actually trying to think about it. As you're going through the actual detailed design of the full product. And especially like, you know, thinking about the firmware requirements, like the firmware engineers, like if they know, they need to build debug code or test code into it at the very beginning they'll architect the firmware completely differently than if it's a afterthought. And one of the issues we run into as well, like with, you know, like all of a sudden firmware engineers are like, oh, we have to test this. We have to build it in last minute and then there's bugs within the firmwares, which can cause, you know, test failures. And so making sure that you are using manufacturing and testing as part of your requirements. Like we need to be able to test the product. Like that should be a requirement. Yeah. Yeah. It's so funny how it's it's so often ignored. That is so often ignored. I, I can't tell you. We had a. We, this is years and years and years ago, probably even before our partnership with CircuitHub, we had a customer come in and there was a purchasing manager who joined, who joined the engineer and they told us we're giving them a tour and everything. And they told us like, in the middle of the tour, it was like, totally like, not even related to what we were talking about. They just said, well, you know, we need you to guarantee that you can deliver a hundred percent functional product to us. And we're like, well, w you don't have a test se like, how can we possibly do that? there's no testing going on. You don't even, they're like, well, we need you to test them. Well, how does the designer of the product, I was a very frustrating experience. So I'm sure that's probably coming across in my voice here, but it's honestly, it was forgivable ignorance because this person just didn't understand. Like they didn't know. They like, you know, we're not making Teddy bears here where it's squishy. Okay. Send it out the door. You know what I mean? Like this is so much more complicated. You really need to think about this stuff in advance. Yeah, it's really important. And remember that, like, I mean, I was definitely guilty of this, of like, oh, like the cm should figure this out and it's like, no, like we're all partners in this. We're all working together. So have an open dialogue and if you don't know, like, you know, you need testing, like don't just like throw it on them. Work with your CM cause like they've tested other products before. So you can ask for a kind of insight on, you know, how you might approach testing a product and stuff like that. So Yeah, don't ask the cm because as we've already learned, I'm too ignorant about it. Just go to fixture, just, FixturFab. Yeah. Just come to us. and and then, and then we'll, we'll work together. All three of us. We'll figure it out. That's really like at the end of the day, it's just, it's just having these conversations, keeping the ball rolling and having the conversations early, even if it's quick, even if it's just you know, you know, a 10 minute discussion where, you know, Duncan says, yep, make sure you get your test points here and all you want to test that. Let's make sure that we can, you make sure to do this in your design. It will be so much easier for us to do the test and you'll save yourself$2,000 in a test fixture. Cause we won't have to do XYZ, you know? Oh, one more tip for designing circuit board only put test points on one layer. Only you points on. What w what does that mean? Like one, you can't get to the internal layers. So what do you mean by So just put test points on the top or on the bottom and try Oh it's yeah, you of course. you can probe from both sides. It just adds a lot of complexity and costs to the test fixture. So it's like, yes, you can do that. But if you don't have to do it, the test fixture is so much easier to build. It's way cheaper for us. It's way more reliable in the long run. Like just try and put them all on one side, either the top or the bottom of the circuit board. Because the operator can just load it upside down or right side up, and that way there hits the bed of nails nails. That's a great to worry about the mechanical tolerances of aligning to different things. I never would've thought of that. I never would've thought of that. It's like, I remember telling a customer, I'm like, oh, you know, it'd be good if you could put fiducials on your board. And, and they're like, oh yeah, you know, we'll be sure to do that. And then they put a fiduciary on like one side and I'm like, well, you have double-sided surface mounts. So I guess I should have specified, we need fiducials on both sides. That's a great bit of advice. Is there any other gems like that you can, you can share with us? Oh, like, locating features of the board. So like, if you can have like either mounting holes or tooling holes, like just something to help us kind of located within the fixture. you can use the outline of the board, but, that depending on like whether or not you're having the board right. If it's routed, it's easy to use that, but if it's like V cut or something like that it, you know, always a little bit of flashing on it, so, That's right? Yeah. The, the oh, I used to know this right off the top of my head, the plus and minus on a, on a V score. It's quite a bit, it's quite a bit it's usually perfectly fine for 99% of products, but if you're trying to use that to justify it somehow in a test fixture or God forbid in your case. That's, that's rough. Yeah. Route routing is, is going to be plus or minus. Oh my gosh. It's like, I think it's like less than a 10th of a millimeter. It's really, those routers are really accurate that do those things. That's great. Is there anything else before we get to my favorite part of the show? Yeah. Like, as you can tell, like, education is such an important part of this process, and we're trying to like address this on our blog. So if you have Oh, beautiful. about, like, we would love to hear, like, what do you want to learn about testing? And if it's not on our blog already, like give us an email at hello@fixturfab.com and we'd like to write about it. Yeah. We'll get make sure to get obviously there's going to be a link to your, your, your website and our on our show notes here, but they could probably just find a link to the blog right. On the homepage I assume. Yup. yeah. Beautiful. One thing that we did years ago, I think Melissa set this up as a, is a nice, a FAQ for just all kinds of stuff, you know, that that's, that can be really helpful sometimes. Just like, you know, can I consign parts to the factory and. You know, and then we have all kinds of fun discussions where Chris says no, and then the salespeople say yes. Yeah. Those are those things, those sorts of things. But yeah, education's huge. I think anybody who listens to this show, we're preaching to the choir when it comes to educating yourself and, and becoming a better engineer. They're gonna eat this stuff up. They're probably going to read your entire blog tonight. I guarantee it. Cause they're going to want to know all about it, especially if they're, if they have to get their products tested. Oh, I had one more question. Okay. So you, you kind of hinted to this. What if your product is like, okay, I got an apple watch on how, how the heck do you test something? A circuit board? This small. I, I have to imagine you don't you just get it into the product and see if it works right. You always try and test at least some part of it. And like, if it's really small and hard to locate, like you can always test it as like a panel. And so like, you'll just have the array of boards prior to them getting broken apart. And so you can just test them all at once. but yeah, with small boards, it can be really hard to locate them within a fixture. Yeah. But you know, at the same time, especially with, don't work so well yet. yeah, no one of the things like we've started seeing is with kind of, Silicon manufacturing, like there's a ton of test fixtures and test probes specifically for like testing, like dyes or, chips and stuff like that. So, we've seen some people using those probes to actually test really small circuit boards, but that's definitely like very specialized use case. And very expensive. I have to imagine. It's gotta be crazy expensive. Yeah. And you know, it's, it's not what we specialize in, Yeah. Have you ever seen a wire bond machine run before oh man. So cool. in using amazing. You talk about fascinating machines to watch? Yeah, that was we were doing image sensors on a product that I've worked on and you can no longer buy, like image sensors that like it's all you have to put the Silicon die on your circuit board and then wire bond. It like during the manufacturing process. Now you can't just like buy an IC, Well, wait, what? No, I, I populate image sensors all the time on circuit boards. What I guess the specific image sensor we were abusing. All right. Okay. All right. That's fine. Understood. so it's like one of these things where I was just like, what? Like we have to do this now. And like, it's so cool watching those machines work. That's the only time I've seen one. It seems impossible because they're light, they're literally like welding this little gold wire to the die and then to the lead. And you're like how it doesn't seem like it should be possible. It's really fascinating. It's like, it's like watching people make Swiss watches by hand. You're like, how is that possible? That should not be possible that you can assemble something that tiny and that precise super cool stuff. Okay. This week in a semiconductor wire bonding, we'll, we'll start a new show. Fantastic. Now I think we're on to, unless there's anything, any other any other tips you want to offer, speak now or forever? Hold your peace. We're on to my favorite part of the show, the pet peeve of the week. What do we. you want to go first? Yeah, I'll go for it. So it drives me crazy when you're at like a sink or a soap dispenser and The sensor just doesn't work and you're sitting The worst. your hand under it just wanted to do a good job, washing your hands and like, can't get the towel. Can't get the soap. Can't get the water. It's the worst. Oh my oh Joe, why didn't I think of that? I have a pretty good temperament, but that's one That makes me quite unreasonably upset when yeah, I think we should just go back to manual valves. I don't think there's anything wrong with that. absolutely a thousand percent agree with you. Where did this come from? That everything had to be touchless. I don't know. It was a good idea until the sensors just stop working And then people skip what using soap. why don't they work? I think this is what I don't. not that hard. I would imagine like a sensors have pretty basic thing. Well, so, okay. So my first thought is maybe the, like it's got a cover, a little plastic cover. Maybe the cover gets dirty and maybe it's some kind of a motion sensor, right? Can you make a motion sensor that doesn't require optics? Is there like a, like a ultrasonic motion sensor or something that just detects something moving in the region has, I don't know. I'm trying to think of something. Maybe they just didn't a test, the circuit board inside the sensor. I guess I will say I stayed at a house one time. We at an Airbnb and it was a pretty nice house and they had a foot pedal valve for the kitchen sink. Have you ever used one of these things before? It's the greatest thing. Ever. Cause you just, you just set the faucet to, to whatever temperature is appropriate. Right. And it's just perfect. Whatever's the nice room temperature for washing your hands, you know, and all day you're just, and then you got to do dishes and so you adjust it to make it a little bit warmer and you just, you just keep hitting that pedal. You're not turning it on and off and having to refine the temperature every single time and your hands are completely free. You're not wasting water. Right. So when you're washing dishes, you're, you're, you know, you get it wet and you get your soap and then, and then you take your foot off and your scrub scrub, scrub, scrub, scrub that way. You're not just letting that all go down the drain, you hit the foot pedal and then you rinse it off. Screw the sensors. Let's just do foot pedals. Yeah, that sounds great. Sign me up. Ah, dude, Joe, that was a perfect pet peeve man. Oh my gosh. It is so true. All right. How about you Duncan? Mine is lately it's I live in Seattle and and Ballard, like you just get a lot of flat tires and I have a all wheel drive car. And like the thing that's currently getting me is the fact that none of the tire shops will replace just a single tire. Like everyone wants me to buy four tires to replace my tires with 1000 miles on them because it's an all wheel drive car and it's going to mess it up. Oh, no kidding. Yeah. Yeah, because you're going to have a different, yeah. But like, you gotta want, there's gotta be a tolerance there. Right. There's gotta be something there where like the manufacturer knows, like, okay, if there's a two millimeter difference, then there's gonna be a problem. But if it's just a one millimeter difference, there's no problem. I'm talking about the, the tread depth, the Yeah, the trend. Yeah. That's yeah, I did find out like if I just go to the Subaru dealership, they'll just replace the tire. Just don't go to a tire shop. So, but yeah. that just irked me. Like I went in and like, literally just like walked out fuming, just like, oh, I can't believe this because this is the second time in like two months that I've had to replace a tire. From potholes or something. Just, I know the first one was a razorblade that like punched through the sidewall. And then this last one was like a big screw that just went through it, but it was too big to plug, we could have a whole pet peeve just on tire shops, period. They're just, have you ever been to like a tire shop and left feeling good? Like had a pleasant experience. I've never had a pleasant experience at a tire shop? Oh, I don't think so. No, gosh. It, it is terrible. I don't get it either. I in and you know what it's like, this is what happened to me is I got burned out on the one tire shop. Cause I had, I got frustrated every time I went there. So I'm like, I'm going to this other tire shop and then. I'm like, oh yeah, there's so much better. But then the second time I went and like, oh God, they're just as bad. There's gotta be a great tire. Shop out there somewhere, charge a little extra and make it a great experience. There's a business opportunity. That is what is it? So I grew up like lake Tahoe area and was it Stone's tire I think was where we'd always get our tires swapped for summer and winter. Cause we all had winter tires and they did good. Like they're just local business. And like that's what I thought like tire shops were like. And then when the first time I like bought tires for a car, like in Seattle, I was like, oh, this is different. So moral of the story is find a local tire shop. That's not a big franchise. And then that way there you get a better experience. There's this tire shop up near us. They give everyone flowers when they leave. Hey, now, Yeah. Like fresh flowers Where is this in Greenfield. I'll drive to Greenfield. I think small business decisions like that make up the world of difference seriously, cause tires stink. So they know you need actually a true story, true story. You know how in the two thousands, the Volkswagen beetle was kind of like relaunched their mind. It might've been in the nineties, they relaunched it. Was an accessory you could buy a little flower cup to put on your dash and people would literally put like dirt and flowers in it and they would, you know, grow little flowers cause you got a nice bright windshield and why not? It's a great environment for a little for a little plant. Well, that was a holdover from the sixties or whenever the beetle was more popular. And the re the original idea behind putting flowers in your car was because back then every car stunk, they didn't have catalytic converters. So the, you know, the exhaust smelled horrible. So they want an offset with flowers. Maybe that's what the tire shop got the idea. Huh? I had no idea. Has been fantastic. I, I really, really, really enjoyed this recording. Is there anything else? Did we miss anything is something else you wanted to talk about? I know you've been very generous with your time. I think that's about it. Thanks so much for having us on like this was awesoome. Yeah. We are hiring a application engineer in our Seattle location. If anyone is interested in joining the team, it's a really fun, just real applications that you're developing for us, like across the board. I know, right on. Well, you got a, there's a great, I mean, the listenership in here, people who would listen to this show would probably make a fantastic applications engineer. So hopefully you hopefully something came out of that. Hopefully you do find a good applications engineer. I'm glad to hear you guys are growing. Keep up the great work. Hopefully we can continue to work together and you know, each day, each time we try to work together, I just never have quite enough volume and quite enough budget again, to fix your main. But I want to, I want to be able to do something, but it's been great having you guys on the show and we we really, really appreciate. Yeah. we did too. Thank you for inviting us to be on the show. Absolutely. Well, if anybody ever wants to get in touch with Joe or Dunkin, we can give you their contact information. Obviously you can find it on their website, which is a fixturfab.com and without the E don't make that mistake. yeah. Did you register the website with the We did. And most of the.co dot CO's dot org.com stat nets. Most of the permutations should get you to our website. Perfect. Perfect. So, yeah. Perfect. Fantastic. So good to hear that as always, you're welcome to reach out to us contact@pickplacepodcast.com. You can tweet at us. We are at CircuitHub and at Wassembly. Thanks for listening to the pick place podcast. If you like, what you heard consider following us in your favorite podcast app, and please leave us a review on apple podcasts or wherever you get your podcasts from. Thanks so much everybody.